Traffic signal lamp recognition algorithm deployment method applied to automatic driving chip

By optimizing and training the traffic light detection algorithm network and deploying it on the autonomous driving chip, the problem of low detection accuracy of existing algorithms in complex scenarios is solved, achieving higher real-time inference accuracy and lower latency impact.

CN119723523APending Publication Date: 2025-03-28ANHUI JIANGHUAI AUTOMOBILE GRP CORP LTD
0 Cites 0 Cited by

Patent Information

Application Number
CN202411894415.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-12-20
Publication Date
2025-03-28

AI Technical Summary

Technical Problem

The existing traffic light detection algorithm has low detection accuracy in complex scenarios, making it difficult to meet the requirements of real-time and low power consumption, especially in the case of changes in lighting conditions, bad weather, long-distance targets and occluded targets.

Method used

A traffic light recognition algorithm deployment method is adopted, including selecting the infrastructure of the object detection model, performing network optimization, collecting visual data in preset scenarios, training the model, and deploying the trained model on the autonomous driving chip for performance verification.

Benefits of technology

The traffic light detection algorithm optimized by performance is deployed on the autonomous driving control chip, which improves real-time inference accuracy, reduces the impact of delay, significantly reduces the missed detection rate, is more adaptable and has higher accuracy.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention discloses a traffic signal lamp recognition algorithm deployment method applied to an automatic driving chip, and the main design concept of the method is as follows: firstly, selecting an infrastructure of a target detection model for traffic signal lamp recognition, and then carrying out the network optimization of the infrastructure of the target detection model, and collecting visual data of traffic lights in a plurality of preset scenes, training the network optimized target detection model by using the visual data, and finally deploying the trained target detection model on an automatic driving chip and carrying out performance test. According to the invention, by deploying a traffic signal lamp detection algorithm with optimized performance on the automatic driving control chip, the real-time reasoning precision is improved and the time delay influence is reduced, so that the problems of poor adaptability to complex scenes, low detection accuracy and the like of the traffic signal lamp detection algorithm are effectively solved, and actual measurement verifies that the omission ratio can be remarkably reduced.
Need to check novelty before this filing date? Find Prior Art